To examine Amazon’s shopping algorithm, we examined the listings for 250 bestselling products across a wide range of categories, from electronics to household supplies, over a period of several weeks this summer.
We wrote a software program that simulated a non-Prime Amazon member whose address was in Brooklyn, New York. For each product, our software scraped the main product listing page and the subsequent pages that contained the listings of other vendors offering the same product.
For items that were listed without a shipping price, our software added them to the shopping cart and proceeded to checkout, where the shipping price was displayed. All of the items that lacked a shipping price were sold by Amazon or sellers in the “Fulfilled by Amazon” program.
Once we had shipping prices for all vendors, we re-calculated the combined price plus shipping for each item and re-ranked the items as they would have been ranked if shipping was included for all items.
Each product’s information was collected during a single browsing session, to ensure that prices did not change while we were collecting data.
We did not examine categories such as virtual goods, e-books and movies that do not have competing vendors and shipping costs. For all products, we excluded sales tax from our analysis.
In total, we examined 6,973 vendor listings.
You can check out our data here.